Researchers have long studied the physiology of the human eye by measuring the alignment of ocular surfaces. Similarly, clinicians often look at the misalignment of the surfaces after implanting intraocular lenses in cataract surgery. They typically obtain this information by collecting and analyzing reflections from the surfaces known as the Purkinje images.Although doctors have used this technique in some form for the past 200 years, they generally do not acquire precise information about the misalignments. In the Oct. 30 issue of Optics Express, investigators at Universidad de Murcia in Spain described an instrument for quantitative measurements of the misalignments in living eyes. The instrument could advance clinical ophthalmology as well as basic scientific studies of the human eye.A method for measuring misalignments of the ocular surface has been reported. It is based on the time-tested technique of measuring Purkinje images representing reflections from four ocular interfaces (for example, air-cornea). The updated method acquires a series of images from nine light sources, as opposed to the single-point source traditionally used with the technique, obtaining more robust information about misalignments of the ocular surface. Shown here are Purkinje images of three of the four ocular interfaces (PI, PIII and PIV). Reprinted with permission of Optics Express.Aligning a light source with the optical axis of the eye should produce four well-centered Purkinje images, the result of the light reflecting at four ocular interfaces: air-cornea, cornea-aqueous,aqueous-lens and lens-vitreous. When the human eye fixates to a point, however, these images typically appear misaligned, likely because of one or more of the following: a global eye rotation, lens decentration and lens tilt. Researchers have reported models relating the positions of the Purkinje images to each of these, contributing to knowledge about the physiology of the eye.Misalignments of the ocular surface generally result from one or more of the following: global tilt, lens tilt and lens decentration. The Purkinje images acquired with the described method help to measure these quantitatively, contributing to better understanding of the physiology of the eye, for example.Ophthalmologists rely on Purkinje images to evaluate misalignment of intraocular lenses implanted following cataract surgery. By measuring the positions of two of the images, which mark the optical axis of the eye, they can determine the extent to which the lenses are misaligned.One difficulty in using Purkinje images is obtaining good images of all four of the reflections. The first two — air-cornea and cornea-aqueous — are similar in size and usually overlap. The third — aqueous-lens — is typically larger, while the fourth — lens-vitreous — is generally smaller and inverted with respect to the others. Furthermore, while the first two reflections are very bright, the third and fourth are typically weak.Shown here are visualizations of three of the ocular interfaces (PI, PIII and PIV) in two normal eyes, as well as a schematic showing the position of each Purkinje image. The researchers continue to develop the software to improve image processing with the technique as well as to locate and reconstruct the Purkinje images automatically.The researchers addressed these limitations through the combination of a CCD camera and a good illumination source. Also, noted Pablo Artal, principal investigator of the study, they recorded series of images with the subject fixating at various angles; typically, Purkinje images are acquired with the subject fixating only at a single point. This approach provided more robust images and yielded structural information about the eye, aiding analysis of the images.The prototype instrument used in the study included a semicircular array of nine infrared LEDs, coaligned with a telecentric camera objective made by Edmund Optics of Barrington, N.J., and a CCD camera made by Hamamatsu of Japan. Use of the semicircular source offered several advantages over the single-point source usually used to produce Purkinje images. First, because of the nonsymmetric geometry of the source, images of the third and fourth reflections were inverted with respect to one another — allowing the researchers to identify them more easily. Also, use of the extended sources meant they could acquire Purkinje images even when the reflections were partially obscured by the pupil.The investigators mounted the instrument on a movable base that also included a chin and forehead rest to hold the subject’s head steady during measurements. The center LED was aligned with the CCD lens axis. They began the session by asking the subject to fixate at the center LED, thus aligning the line of sight with the instrument axis. Then they recorded an image of the anterior part of the eye for each of the fixation positions. Finally, using image processing software written in-house, they established the location of each of the Purkinje images with respect to the center of the pupil.They first tested the instrument in subjects in whom intraocular lenses had been implanted during cataract surgery. The benefit of beginning with these subjects was that the reflections were brighter and that the precise geometrical data of the lenses were known. In fact, the researchers obtained very well defined Purkinje images in these eyes.Then they performed measurements in the natural crystalline lens in two normal eyes, in subjects 26 and 50 years old. They anticipated a worse signal-to-noise ratio in the natural lens reflections than in the intraocular lens reflections, in part because the reflection for the third Purkinje image is relatively diffuse. They found, though, that they could readily acquire this image most of the time and that image processing enabled them to determine its position.The software also allowed them to separate the effects of a global eye rotation from those of a lens tilt, with only two constants needed in the analysis. A previously established method requires nine. The latter and other similar methods are less complicated in that the set of Purkinje images is contained within a single image. The researchers believe, however, that the series of images used in the described method provides more robust and more accurate data.The scientists continue to develop the instrument in anticipation of clinical implementation. “The optics is, in fact, very, very simple,” Artal said. “We are now working on the software to locate and reconstruct the Purkinje images automatically. Currently, it’s a semiautomatic process.” He added that this is the primary obstacle to clinical implementation, as use of the device still requires a significant understanding of how it works. They hope to have completed a second generation of the instrument, with full automation, within a few months.